stretched / rates / UST 2Y · data through 2026-09-15
UST 2Y
STRETCHED · CROWD LONG · WEEK 7★ FREE PREVIEW · every Pro file looks like this☆ Watch this marketFRIDAY ALERT · SIGN INBig speculators have been betting on UST 2Y rising for 7 weeks, a bigger bet than in 99% of the last 3 years. When a market gets this one-sided, it sometimes snaps back hard. Below: what happened every other time it looked like this.
the book: 704,048 long vs 1,559,401 short · this week specs added 125,091 longs · added 51,337 shorts
STEP 1 · THE REPORT & THE CHART
Who holds what in UST 2Y
The COT report you already read, on a chart. Free.
STEP 2 · THE BACKTEST
Every past extreme, numbered
What turned, what didn’t, what each one cost to hold. Free.
STEP 3 · THE CONFIRMATION LAYER
You bring the entry. This is what sits on top of it: how far along this setup is, what marked the turns here before, how much room it needed, and how long the wait ran.
HOW OFTEN A CROWD THIS SIZE ACTUALLY TURNED
THE ENTRY METER · 1 OF 5 STAGES LIT
reads 3× a day · updated 2026-09-23
LATEST EVENT · quiet on the tape · last read 2026-09-23
● 1 · CROWDED
99th pctile · week 7
○ 2 · CONDITIONS
no separating markers here
○ 3 · CROWD BLINKS · LIVE
the crowd is holding its bet
○ 4 · THE FIRST CLOSE
not here yet · no close against the crowd
○ 5 · THE WINDOW
starts if we open a fade here
even fully lit, this is a risk state · across comparable extremes 38 in 100 turned within 8 weeks (ordinary week: 37) · the ladder tells you where you are, never what happens
Nothing we track told the turns apart from the duds here, across 220 turns. That is the measurement, not a gap in it.
Nothing here reliably told turns apart from duds. That is the measurement, not a gap in it.
HOW THIS WAS MEASURED + −
‘Mover’ = median/share among past episodes that reversed within 13 weeks (pooled across all markets · this market’s sample is thin). Only parameters whose two groups actually differ are printed · the rest read the same at the turns as when nothing turned, so they carry nothing. A parameter is scored on which of the two medians today sits nearer, which is a distance, not a probability. A profile, not a rule · your judgment decides.
86/100
HIGH PRESSURE
CROWD PRESSURE · as of 2026-09-23
99
LEAN
70
WEEKS
—
FINGERPRINT
52
TAPE
how hard the crowd leans, how long, how much of the turn fingerprint is on today’s tape, and whether it is still adding · a pressure reading, not a forecast
THE DIAL, GRADED: the same formula, scored at entry on every past case · pressure ≥70 turned ≤8w in 65/186 (35%) · 40–69 in 410/1,011 (41%) · all cases 475/1,197 (40%) · pressure measures how big and stubborn the crowd is, not when it breaks: it is not a turn probability · the timing layer is the marks below
The weeks that looked most like this one
MEDIAN AFTERMATH · -1.8% IN 13 WEEKSThe five weeks that looked most like today went -1.8% over the next 13 weeks, in the middle.
HOW THIS WAS MEASURED + −
Jan 2026
crowd 72 of 100
-0.3%
next 13w
Nov 2021
crowd 88 of 100
-1.8%
next 13w
Jul 2022
crowd 77 of 100
-2.1%
next 13w
May 2022
crowd 85 of 100
-1.3%
next 13w
Jan 2022
crowd 94 of 100
-2.8%
next 13w
a measurement of similarity, never a forecast · similarity is scored on positioning and the shape of price against this market’s own record, and what followed is the plain weekly price change
the ringed dot is your whole edge.
The marker that showed up most often at past turns here appeared at 100% of them, typically 2 days after the high or low. We check it every day and tell you the moment it fires. It also fires in ordinary weeks, so it only counts while the crowd is this stretched.
What marked the past turns, we check these daily
SCOPE: UST 2Y'S OWN 14 TURNSThese are the signs that showed up first when this market turned before. We watch for them every day, and tell you the moment one fires.
HOW THIS WAS MEASURED + −
First close beyond the 5-day level, against the crowd
typically 2 days after the top/bottom · 81% of the move still ahead · ~57 false alarms a year
First close beyond the 10-day level
typically 5 days after the top/bottom · 72% of the move still ahead · ~40 false alarms a year
Specs cut in Friday’s report
typically 5 days after the top/bottom · 63% of the move still ahead
Open interest drains ≥2% in a week
typically 8 days after the top/bottom · 75% of the move still ahead
Tradeable from 2026-09-21. Size for the worst case, not the average.
You just read the full file. Free, no card.
Pro opens this exact file on every crowded market, every week · cancel anytime
The Friday Letter
Every Friday 21:30 CET the new report lands and we regrade all 1,206 cases overnight. Confirm and your crowd-map snapshot arrives right away. Then every Friday: what moved, where the crowd is stretched, and which fades are running in public. And the moment we open one, you get the ping. Free.
Every Friday: where the crowd is stretched, what is running, what it cost. Free.
Confirmation by e-mail, one-click unsubscribe in every issue. Your address is never shared.
These are statistics about the past, not advice. Most comparable setups did NOT turn. A reading here means the odds have historically been better than normal, never certain. Our data and models can contain errors; past extremes do not predict future turns, and futures losses can exceed your deposit. The decision, and its full risk, is yours.